A Manual for Practitioners

Scott R.Pearson Eric A. Monke

_ _ " _ ......

/-. *1

MkO

FORM. .....

Contract No. PDC-0091-C-00-6176-00

THE POLICY ANALYSIS MATRIX A Manual for Practitioners

Scott R. Pearson Eric A. Monke

Submitted to: Office of Policy Development and Program Review

Bureau for Program and Policy Coordination U.S. Agency for International Development

Washington, DC 20523

Submitted by: The Pragma Corporation 116 East Broad Street

Falls Church, VA 22046

July 1987

Scott R. Pearson is a professor in the Food Research

Institute, Stanford University, and Eric A. Monke is an associate

professor in the Department of Agricultural Economics, University

Gf Arizona. The authors owe thanks to Mark Langworthy, Assistant

Professor, Department of Agricultural Economics, University of

Arizona, for assistance in developing computer software and for

writing Annex ?", and to Melanie Sander:-, Project Manager, The

Pragma Corporation, for administrative and editorial assistance.

CONTENTS

LIST OF TABLES AND ILLUSTRATIONS Vii LIST OF BOXES viii EXECUTIVE SUMMARY ix

I INTRODUCT ION KO THill.B 1MANUAL

A Framework for Agricultural Policy Analysis 2 Methods of Po licy Analysis 6 Practi.cal 15sw s Addressed 8 Structure of the Manual 10

2 INTRODUCTION TO Till POLICY ANALYSIS MATRIX 13

Policy Analys is Matrix 13 Comparisons Among Agricultural Systems 22 ResearchI Plan for Efficiency and Policy Analysis 26 Costs o Carry ing Out the PAM Approach 28

30Scmwvuam

3 CONSTRUCTI ONG 'AMs FOR COMMODITY SYSTEMS 33

Defining the Commodity System 33 Classification of Inputs and Outputs 36 Evaluation of inputs and Outputs 44 Disaggregating Input Costs 46 Activity Bud]ets and the CommodLty System 50

4 FARM-LEVEL BUDGETS AND ANAIYSIS 63

Selection of Representative Crop Systems 64 Procedures for Budget Preparation 65 Collection of 1Inout and Output Data 68 Complex (Cmmdity 5':s ems 74

5 POST-FARM BUDGE.TS AND ANAiNSIS 87

Sources of Variation in Representetive Activities 88 Identificat.ion and Selection of Representative Activities 90 Procedures for Data Col.ection 91 Survey Strategies 98

v

http:BUDGE.TS

6 ESTIMATING SOCIAL PRICES FOR THE PAM 113

Theoretical Foundation for Social Price Estimation 114 Estimating Social Pr,-ices for Tradabies 119 Estimating Socil P:ices foc Domestic Factors 127 Sensitivity Analysis 136

7 INTERPRETATION OF R.ESULTS FOR POLICY ANALYSIS 149

Interpretation of the lifects of Divergences 150 The Policy Analysis Matrix and Public Policy 160

8 COMMUNICATI.NG RESULTS TO POLICYMAKERS 187

Practical Suggestions for Effective Writing of Policy Memos 189 Practical Suggestions for Effective Oral Presentations 194 Designing Research and Communicating Results 198

APPENDIX A USING THE PAM ON THE MICROCOMPUTER 205

vi

http:COMMUNICATI.NG

TABLES

2.1 Policy Analysis Matrix 14 2.2 Expanded Policy Analysis Matrix 20 7.7 Farm-Level Profitability and Land Rent by Soil Type and

Crop, 1983 and 1996 (in Thousands of Escudos/Hectare) 182 7.8a Social Returns, Costs, and Profits or Alternative

Locations and Milling Techniques 184 7.8b Net Savings Over Manual Cultivation from Changes in

Techniques, Inland Countries 186 8.1 Essential Elements of Policy Memos 199

FIGURES

3.1 The Structure of the Coricdity System fur PAM Analysis 37 3.2 Input and Output Categoviies for Activity Budgets 38 3.3 The Structure of Activity Budgets for the PAM 47 3.4 Conversion Ratios and the Calculation of System Costs

Returns 53 5.1 Marketing Margins and Storage Costs 95 6.1 The Determination of Social Output and Input Prices in 115

a Simple Economy 6.2 Expanding the PAM to Incorporate Costs for Land 135

vii

BOXES

2.1 Numerial Examples of Policy Analysis Matrices 31 3.1 Selecting a Representative Wheat Production System 54 3.2 Calculating Annual Equivalent Values of Fixed Inputs 56 3.3 Decomposition of Nontradable Inputs 58 4.1 Indentification of Commodity Systems 81 4.2 Inputs and Outputs in the Farm Activity Budgets 83 4.3 Calculation of the PAM for Permanent Crops 84 5.1 Alternative Marketing Chains for Delivering Rice to

Urban Market, Atebubu District, Ghana 105 5.2 Processing Budget for Small-Scale Processing Operations--

RAMAS Flour Mills in Portugal 107 5.3 Sample Questionnaire for Large-Scale Processing Mills 108 5.4 Partial Budget Adjustments for Full Utilization of Flour

Mill Capacity in Portugal 110 6.1 Social Price of Rice 139 6.2 Rate of Return to Capital 142 6.3 Social Wage Rate 145 6.4 Social Value of Land 147 7.1 Output Transfer in Portuguese Wheat System 166 7.2 Tradable Input Transfers in a Portuguese Wheat System 168

Wheat System

7.3 Effective Protection Coeffecient for a Portuguese

170 7.4 Factor Transfers in a Portuguese Wheat System 172 7.5 Net Transfers, Profitability Coefficient, and Subsidy

Ratios to Producers for a Portuguese Wheat System 174 7.6 Price Policy Graph for Rice in Indonesia 176 7.7 Projected Impact of Price Policy Changes in the Private

Profitability of Portuguese Agricultural Systems 180

Production in Four West African Countries

7.8 Profitability and Technological Change in Rice

183 8.1 Illustration of Elements of a Policy Memo 201

viii

EXECUTIVE SUMMARY

Agricultural policy consists of decisions that influence the

levels and stability of crop and farm input prices, the choice of

public investments affecting agricultural revenues and costs, and

the allocation of public research funds that are used to improve

farming and processing technologies. The purposes of this manual

are to suggest a simplified approach to agricultural policy

analysis, to explain the empirical method of analysis in full

detail, and to describe how the results might be interpreted and

communicated effectively to policymakers.

The method described in this manual is the Policy Analysis

Matrix (PAM). The PAM approach serves both as a logical

framew,rk for thinking about the effects of agricultural policy

and as a method of empirical analysis for measuring such policy

effects. It is designed to permit. easy communication of methods

and results between policy analysts and policymakers.

Application of the PAM approach allows analysts to address

questions of commodity price, public investment, and agricultural

research policy.

An agricultural system is a chain of farming, marketing, and

processing activities that together produce a marketable product,

such as wheat flour or milled rice. Every country has a number

of such systems that are differentiated by commodity, technology

of farming or processing, and agro-climatic zone. During any

given year, each of these systems earns revenues from sales,

incurs costs of production, and receives profits if revenues

exceed costs.

The first task in applying the PAM approach is to select the

principal agricultural systems, collect farm and post-farm budget

data on revenues and costs in some base year, and calculate the

level of positive or negative profits for each system. Because

these revenues, costs, and profits are all based on actual market

data, the first row of each PAM, into which they are placed, is

termed the "private" row. Private profits are an important

result because they indicate the degree of competitiveness of the

ix

system; they can also be broken down to show farm, marketing, and

processing incomes separately. Entrepreneurs will pursue

activities continuously only if they yield private profits.

Economic planners need to know more than whether

agricultural systems are privately profitable and by how much.

Those responsible for agricultural policy are concerned with

furthering government objectives. Foremost among these goals is

the efficient use of the country's scarce resources to achieve

highest levels and growth of national income and output. Other

objectives, such as income distribution and security of food

supplies, are often furthered oiily at the cost of a loss in

efficiency; policymakers then have to tradeoff the value of the

perceived nonefficiency benefit against the measured efficiency

loss. Measurement of efficiency of agricultural systems,

therefore, is essential even when the government chooses to place

higher weights on nonefficiency objectives. Use of the PAM

approach provides this efficiency measure.

The second row of each PAM matrix is the efficiency or

"social" row. Each agricultural system's revenues and costs are

valued in hypothetical efficiency or social prices that are meant

to bring about the most efficient allocation of the country's

resources in production and thus lead to maximal output and

income. The task of efficiency pricing in the PAM approach is

the same one faced by planners doing social benefit-cost analysis

of investment projects. A central intention of this manual is to

simplify this often mysterious task so that analysts and

decisionmakers can feel confident in assessing social prices.

The first principle of efficiency pricing is that comparable

"world" (cif import or fob export) prices serve as social

valuations of commodities or inputs that are traded

internationally. If product markets are operating reasonably

efficiently, world prices would determine domestic price levels

in the absence of government policy. But by using tariffs,

quotas, taxes, or subsidies to influence levels of imports or

exports, governments can cause their domestic prices to be higher

x

or lower than the comparable world prices. Changing the domestic

price level and degree of instability relative to what they would have been if determined by world prices is the essence of agricultural price policy. But unless price policy is done to

offset some failure of a market to operate efficiently, this

action distorts domestic prices away from their most efficient

levels, which are given by the world prices. Social valuations

of outputs and of tradable inputs are thus the world prices for

these items, adjusted for location and quality.

Social valuation would be straightforward if all outputs and

inputs had world prices, but not all do. Domestic factors of production--labor, capital, and land--are not usually tradable

internationally, and their prices are instead determined in

national markets. Social valuation of domestic factors can be a

complicated exercise. The idea is to find the social. opportunity

cost of each factor, the amount of nationil a'icome foregone trom not using the factor in its next best alternative use. This is done by making adjustments to the observed market price that

would eliminate any distorting effects of output price policies

and of factor market policies. Intermediate inputs that do not

enter international trade are broken down into their component

costs of tradable inputs and domestic factors to permit social

valuation.

Valuation of social profits is a principal result of the PAM approach. Positive social profit indicates that the system

operates efficiently and that the country has a comparative

advantage in producing the products using that system's

technology and agro--climatic zone. Negative social profits mean

the opposite; distorting policy allows the system to operate with

positive private profits, but the country is losing potent.Lai income by using resources to produce the given products with the

indicated technology and agro-climatic zone.

A system producing with negative social profits is at a strong disadvantage as a candidate for receiving new allocations for agricultural research or new public investment projects. The

xi

evaluation of benefits from both of these dimensions of

agricultural policy properly counts only additions to positive

social profits. Any "catch-up" to make currently unprofitable

systems at least break even in social prices cannot be counted in

assessing net benefits of public spending. For this reason, it

is useful to carry out PAM analyses of principal agricultural systems to establish baseline information on social

profitabilities; these results aid the allocaticn of public funds

in the agricultural sector among competing investment

opportunities and to alternative research directions.

The third and bottom row of each PAM matrix contains entries

that ixdicate the difference between private and social

valuaticons, which are defined as the effects of "divergences." A

divera.!nce arises when government policy creates an inefficient

allocation of resources ("distorting policy") or when a commodity

or factor market operates inefficiently ("market failure"). An

"efficient policy" is one that offsets a market failure.

Virtually all the effects of divergences for tradable outputs and

inputs typically are caused by distorting policy and can thus be

removed by decisions policy changes. lor factors, the effects of

divergences usually are a composite of distorting policies and

market failures. If market failures are unimportant, the bottom

row measures mostly the influences of government policies on

agricultural commodities and inputs, on exchange rates, and on

domestic factors. Use of the PAM approach thus permits

measurement of both efficiency and policy effects for the

country's principal agricultural systems.

Effective agricultural policy analysis consists of four

steps--matching the policy issues with appropriate analytical

methods, carrying out the analysis, interpretizg the results, and

communicating the meaning of the analysis for policy choice to

decisionmakers. The first six chapters are concerned with the

first two of these steps. Interpretation of the results in the

context of policy systems is the focus of Chapter 7, and

effective communication of the lessons to policymakers, in policy

xii

memos and in oral reports, is the topic of Chapter 8. Because the PAM approach is designed both to provide a conceptual path to policy analysis and a method of measuring policy transfers, PAM results are easily communicated to policymakers.

xiii

CHAPTER 1

INTRODUCTION TO THE MANUAL

How can decisionmakers in developing countries easily

understand the individual and cumulative effects of government

policies that influence the competiLiveness of agricultural

production systems? How, for example, might the combined effects

on the profitability of taiming rice of a quota on imports of

rice, a subsidy on fertilizer, and an overvalued exchange rate be explained to ministers in simplified yet accurate terms? What

are the principal relationships between such policies and the

economic efficiency of agricultural production, and how might

these linkages be readily communicated to domestic investment

planners and foreign aid officials? The purposes of this manual

are to suggest a simplified approach to agricultural policy

analysis, which can address these questions and others like them, to describe the empirical approach in detail, and to explain how

the results might best be interpreted and communicated to those

who influence or decide agricultural policy.

This document is designed to be a manual, a st-aightforward

explanation of how to do one kind of applied economic analysis.

The manual is addressed to prdctitioners--economic analysts in

planning agencies of agricultural and related ministries or in

foreign donor Lffices--whose jobs are to analyze economic aspects

of agricultural production systems. The role of these

practitioners is thus to conduct agri-ultural economic

analysis--for example, to study the competitiveness of different

cropping systems, distinguished by commodity mix, technology, and agro-climatic region; to investigate how public investments might

change the efficiency of those systems; and to aid researchers in

selecting technologies that will improve competitiveness and

efficiency.

All ecenomic analysis is based on methodology, a combination

of definitional identities, behavioral parameters, and arbitrary

assumptions, which should be grounded firmly in economic logic.

2

A problem arises for practitioners of any method, however, if the

designers of the approach fail to explain in adequate detail

exactly what is involved in conducting the analysis, what the

costs and benefits of the research are likely to be, and how the

results can be easily explained to those who ultimately set

policy or allocate funds. The purpose of a manual is to fill

these needs--to explain what has to be done to apply a method,

when to invest the time and money required in the research

effort, and how the results can lead to better informed

decisions.

A frustration for many economic analysts is that few such

manuals exist. What are available, instead, are complex

methodological sections, chapters, or appendices that are

understood only by a small set of experts who specialize in

refining the approaches. Practitioners, including those in

developing countries, are usually left either to puzzle che

mysteries of the unused approach or to plunge in and make serious

errors as they attempt to learn by doing. This manual is an

attempt to fill this gap for one method that addresses several

critical issues of agricultural policy analysis for developing

countries.

A Framework for Agricultural Policy Analysis

A useful construct for approaching the analysis of

agricultural (or any) policy is the objectives-constraints

policies framework. This framework serves principally as a

mindset for organizing one's thinking about policy evaluation.

Objectives

Governments are assumed to have broad objectives that they

are trying to further through interventions in the agricultural

sectcr. The three most common objectives are efficiency--the

allocation of resources to effect maximal nationI output and

income, equity--the distribution of agricultural incomes to

preferred income groups or regions, and food security--the

short-run stability of food prices at affordable levels to

3

consumers, reflecting adequacy of food supplies, and the long-run

guarantee of adequate human nutrition. Any government action that can simultaneously further all three objectives should be taken. Typically, however, the promotion of one objective conflicts with one or both of the others. In this situation, policymakers need to tr$ deoff the gains in one area against the

losses in the others. For example, losses in efficiency, if not too large, might be tolerated if the action were believed to result in significant improvements in income distribution or food security. Decisions between competing objectives is the essence of policy analysis. Policymakers do it, explicitly or

implicitly, by making value judgments about the value of furthering different objectives.

Constraints

The need to make these often difficult choices arises because of constraints in the economic system. Three categories

of constraints limit the agricultural sector from realizing its potential. The ability of the country's agricultural systems to produce commodities is limited by supply constraints--the

availability of domestic resources (land, water, labor, capita),

the existence of production technologies (for farming and

processing), and the relative costs of all inputs. The value of the commodities produced is in part established by domestic

demand constI.dints--levels and growth rates of populations and

incomes; changes in taste preferences, and the relative prices of agricultural commodities. Both domestic supply and demand

constraints are moderated by the world prices of agricultural

outputs and inputs that enter international trade. Because these world prices, the third constraint, determine the domestic prices

of internationally tradable commodities when no policies

intervene, all agricultural price policies either increase,

decrease, or stabilize domesFkc prices relative to the underlying

world prices. The responsAveness of producers and consumers to

these price policies depends on the underlying supply and demand

constraints, which in turn condition producer and consumer

4

behavior and thus the Ehape and position of supply and demand

schedules. For each agricultural system, therefore, the three

categories of constraints can be depicted by drawing a supply

curvu, a demand curve, and the relevant world price line for the

outputs (the cif import price for goods that are partly imported

or the fob export price for commodities some of which is

exported).

Policies

Policies are the instruments of action that governments

employ to affect change. Three principal categories of policies

are used to bring about change in agriculture. The first, illustrated above, is agricultural price policy. Two hain types of price policy instruments can be used to alter prices of

agricultural, outputs or inputs. Quotas, tariffs, or subsidies on

imports and quotas, taxes, or subsidies on exports directly

decrease or increase amounts traded internationally and thus

alter domestic prices; these trade policies apply only to volumes

traded internationally and not to domestic production. Domestic

taxes or subsidies, in contrast, create transfers between the

govern;ient treasury and domestic producers or consumers; some

cause a divergence between domestic and world prices, but others

do not.

The second category of policies is nationwide in coverage.

Macroeconomic policies comprise the central government's

decisions to tax and spend (fiscal policy), to expand the supply

of money (monetary policy) ; to influence the foreign exchange

rate (exchange rate policy) , and to intervene in the markets

where the prices of the primary factors (wage, interest, and land

rental rates) are determined. Although these policy decisions

are not typically taken principally because of their impact on

the agricultural sector, they have a very important impact on

agricultural systems. Macro policies can affect both output and

input prices in agriculture. Sometimes the macro policy effects,

however unintended they might be, more than offset the desired

incentives of agricultural price policy.

5

In addition to price and macro policies, governments

influence their agricultural sector through public investment policy. Government budgetary resources can be invested in

agriculture to increase productivity and reduce costs. The most common ways of doing this include investments ir. agricultural research to develop new technologies with higher yields or lower costs, in infrastructure (roads, irrigation, ports, and mar):eting facilities) , in specific agricultural proj ects to increase production and to demonstrate new technologies, and in educating and training agriculturists to upgrade the human capital in the

sector.

Summary

With its three components in place, the framework for policy

analysis can be summarized easily. Policymakers enact policies

(price, macro, or investment) to further government objectives (efficiency, equity, or food security) in the face of economic constraints (supply, demand, and world prices). Policy analysis ';onsists of evaluating price, macro, or investment policy instruments by quantifying the cons:raints and by estimating the likely impacts of policy on objectives. Analysts can thus identify tradeoffs between objectives and attempt to measure their magnitudes. Pol-icymakers can thcn hetter exercise their value judgments about what is desir able policy.

By placing differentt weights on the importance of objectives, policymakers c an justify aImost any government action. The main services that: economic analysis can provide to policymakers are to distinguish whether a policy is likely to improve the efficient operation of the economy and thus assist faster growth of national income (the most widely used indicator of successful economic management) ; to measure the expected magnitude of the efficiency gains or losses; and to quantify, when possible, the direction and extent of the policy's likely effects on the equity and food security objectives. Even when the nonefficiency effects are difficult to measure, economic analysis can provide a reasonable estimate of any short- and

6

long-run efficiency costs likely to be associated with promotion

of nonefficiency objectives.

Methods of Policy Analysis

In an ideal. world, policy analysts would have unlimited time

and resources to evaluate policies. In practice, however, policy

analysts face the binding constraints of limited time and

resources. Policy evaluations and decisions are made in short

time periods, and decisions are usually taken whether or not economic analysis is complete. In most developing countries,

policy evaluations are made by a small number of professionals

with only limited support from research assistants and data

processing equipment. Data availabilities are often constrained

as well. In many developing countries, agricultural sector data

are scarce and of dubious quality--data on supply and demand

functions, possibilities for input substitution and technical

change, or costs and returns for the production of alternative

crops are often not available to permit an analysis of policy

effects. Finally, the analyst will usually have only limited

access to policymakers. Hence, the analyst is unable a priori to

measure the relative importance of various objectives and thus

can only escablish the contribution of policy to particular

objectives, but not the "appropriate" tradeoff among them. In

many instances, the "appropriate" tradeoff is probably not known

explicitly by policymakers either. Instead, trade-offs and

weights in objectives emerge from ex post discussion and

arguments over the various effects of policy.

Traditionally, economic analysis of agricultural policy has relied heavily on the estimation of supply curves for various

inputs and outputs. In principle, these estimates should provide

an accurate assessment of market behavior and response. But, in

practice, sufficient historical data of reliable quality are only

rarely available to permit an analyst to assess fully the impacts

of a complete set of government policies on the behavior of a

particular agricultural system; elasticities of supply

7

(statistically estimated parameters that relates the percentage

change in product price to a percentage change in quantity

produced) require long time series of data, usually 25 years or

longer. Even when private marginal costs curves can be developed

for output responses, input demands and the impact of various

interventions on production costs are necessarily overlooked.

The resulting policy analysis remains incomplete and often

incomprehensible to policymakers. Resolutions to these problems

are being sought. Both domestic and international organizations

are continually engaged in improving systems and methods of data

collection. Advances in computational equipment and analytical

methods are improving the prospects for understanding better the economic responses to agricultural producers. But these advances

have not yet provided a satisfactory set of methods and

parameters for policy analyses when reliable historical

information is unavailable.

These constraints imply that agricultural policy research

ought to use a more simplified, pragmatic, and disaggregated

analytical framework. Ideally, the framewuick should yield

results that are simultaneously comprehensible to policymakers

(who might have limited understanding or respect for economic

analysis or policy) and yet retain theoretical consistency. The

approach advanced in this manual builds around a framework termed

the policy analysis matrix (PAM) . Based on an assessment of average costs rather than marginal costs, the method contains

numerous theoretical assumptions and si mplif ications, and a thorough understanding of these limitations is essential for

useful application of the method. But in most situations, the

advantages of the method outweigh its shortcomings. Requisite

data on farming, marketing, and processing budgets are available

or can be quite easily collected, and evaluation can proceed in a

timely manner. Iin principle, valuations of both private incentives and the effects of policy on agricultural systems are

feasible. Most important, the method allows measurements of the

effects of policy on producer wel fare as well as the

8

identification of transfers among interest groups--producers in agricultural systems, consumers or food, and policymakers

controlling allocations of the government budget. These items form a critical set of inputs in any evaluation of agricultural

policy.

Practical Issues Addressed

The method described in this manual is the Policy Analysis

Matrix (PAM). The PAM approach is introduced in the following chapter, and its theoretical underpinnings, including its links to international trade theory and to the theory of social

benefit-cost analysis, are spelled out fully in an unpublished book manuscript by the authors entitled The Policy Analysis

Matrix and the Evaluation of Agricultural Choices. Three principal categories of issues--the impact of agricultural price

policy on competitiveness and farm-level profits, the influence

of investment policy on economic efficiency and comparative

advantage, and the effects of agricultural research policy on changing technologies--can be investigated with the PAM approach.

Competitiveness and Farm Profits

What kinds of farmers--categorized by the commodities they grow, the technologies they use, and the agro-climatic zones in

which their farms are located--are competitive under current

policies, and how would their profits change as the price policies are altered? This central issue of farm policy--how

agricultural prices affect farming profits--is of primary

importance in ministries of agriculture that represent farm

interests. In the PAM approach, farm budget data (sales revenues and input costs) are collected for the principal agricultural

systems. Calculation of profits actually received by farmers is

a straightforward but important initial result of the analysis.

It shows which farmers are currently competitive and how their profits might alter if price policies were changed. A major

advantage of the PAM method is that much of the information it

requires is already available or readily obtainable in

agricultural ministries or statistical agencies.

9

Efficiency and Public Investment

A second issue, addressed by the PAM approach, concerns the

existing economic efficiency (or 2omparative advantage) of

agricultural systems, defined as linked chains of farming,

processing, and wholesale marketing activities, and how

additional public investment might change the current pattern of

efficiencies. In what commodity production systems, defined by

technology and agro-climatic zone, does the country currently

exhibit strong or weak comparative advantage, and how might new

investments, using government revenues or foreign aid funds,

increase efficiency? This critical issue of investment policy is

of primary interest to economic planners who allocate the

country's capital budget, including foreign aid, in an attempt to

increase efficiency and speed growth of national income.

With the PAM method, the analyst reassesses the revenues,

costs, and profits obtained by collecting farm-level and

post-farm budgets. This is achieved by calculating efficiency

valuations of outputs and inputs--which are meant to lead to the

highest possible levels of national inccme--and then using these

"social" valuations to measure the value of the commodities

produced and of the inputs used in their production. The

difference between revenues and costs fcr a system, both valued

in social prices, gives social profits, which is a measure of

economic efficiency. If new investmen:s reduce social costs,

they also increase social profits and improve efficiency. For

well-informed investment planning, it is essential to have a good

understanding of the current array of social profitabilities of

agricultural systems, because this information reduces greatly

the number of detailed benefit-cost analyses needed to evaluate

investment alternatives.

Efficiency and Agricultural Research

A closely related issue is how best to allocate funds for

agricultural research. How can economic analysis be used to

identify the most fruitful directions for primary and applied

research that aim at raising crop yields, reducing social costs,

10

and thus increasing social profits? This question is faced by decisionmakers in the international agricultural research centers

(the CGIAR, headquartered in the World Bank), in several other

international organizations, and in the agricultural research establishments controlled by single countries. It is one also asked by central planners who make allocations to agricultural

research budgets.

The answer, available through PAM, is analogous to that for

choices among public investment projects. The existing levels of private (actual market) and social (efficiency) revenues, costs,

and hence profits need to be estimated for principal agricultural

systems. This calculation will show the extent to which actual profits are due to policy transfers and not the result of

underlying economic efficiency. Neyt, the agricultural scientists need to project the yield-increasing and cost-reducing

changes they expect to result from alternative research programs.

The effectiveness of such changes can then be gauged by examining

how they alter private and social rrofits, starting with current

technologies and policies.

Structure of ti~e Manual

Because this is a manual for practitioners, the practical or

how-to-do-it aspects of the PAM approach are emphasized

throughout. The manual, therefore, introduces essential

principles and examples that illustrate how to resolve or prevent

problems. Because good applied analysis is academic if its

results are not clearly communicated to poLicymakers, the

practical focus of this manual extends to interpretation and

explanation of results.

The structure of the manual follows the logic introduced in

this opening chapter--first principles, then procedures, and

finally interpretations. Chapters 2 and 3 deal with the

principles of PAM and of constructing budgets; the purposes are to introduce the logic of the method and to illustrate how one

organizes the basic data. Chapters 4, 5, and 6 lay out how to

ii

build farm-level and post-farm budgets and how to do private and social valuations; these chapters contain the essence of applying

the PAM method and of generating results. Chapter 7 discusses

how to interpret the results completely and responsibly. Chapter 8 contains suggestions on communicating the results in writing

and orally, to busy policymakers. Appendix A demonstrates how a

microcomputer can assist the analysis, although the procedures

can be done, fully if tediously, with a hand calculator.

CHAPTER 2

INTRODUCTION TO THE POLICY ANALYSIS MATRIX

This chapter explains the construction of the Policy

Analysis Matrix (PAM) and the derivation of the measures of

efficiency used in agricultural policy analysis. The main task

is to construct accounting matrices of revenues, costs, and

profits. A separate PAM is constructed for the study of each

selected agricultural system--comprising data for a single

commodity from budgets on farming, farm-to-processor marketing,

processing, and processor-to-wholesale center marketing. In any

given year the impact of commodity and macroeconomic policies can

then be gauged through comparison with the situation in the

absence of such policy. The chapter concludes with a discussion

of a research plan for PAM.

Policy Analysis Matrix

PAM is a product of two accounting identities--one defining

profitability as the difference between revenues and costs, and

the other measuring the effects of divergences (distorting

policies and market failures) as the difference between observed

parameters and Parameter levels that might exist if the

divergences were removed. By completing a PAM for an

agricultural system, an analyst can simultaneously measure both

the extent of transfers occasioned by the entire set of policies

acting on the system and the degree of economic efficiency of the

system.

Profitability is a basic concept of economic analysis.

Profits are defined as the difference between total (or per unit)

sales revenues and costs of production. This definition of

profitability is the first identity of the accounting matrix. In

PAM, profitability is measured horizontally, across the columns

of the matrix (see Table 2.1). Profits are found by subtracting

costs from revenues. Each column entry is thus a component of

the profits identity; revenues less costs equals profits.

14

Private Profitability

The data entered in the first row permit a measure of

private profitability. In the PAM method, the term private

refers to observed data on revenues and costs, reflecting actual market prices received or paid by farmers, merchants, or processors in the agricultural system. The private or actual

market prices thus incorporate the underlying economic costs and valuations plus the effects of all policies and market failures

that create transfers in the system. The first step in the empirical application cf PAM is to calculate the private

profitability of an agricultural systiem in some base year,

usually the most recent year for which detailed data are available. In Table 2.1 private profits, defined as D, are shown

as the difference between revenues (A) and costs (B+C), and

Table 2.1 Policy Analysis Matrix

Revenues Costs Profits Tradable Domestic

input factors

Private prices A Social prices E Effects of divergences 13 and efficient policy

B F J4

C G K5

D1 H2 L6

1 Private profits, D, equals A minus B minus C.

2 Social profits, H, equals E minus F minus G.

3 Output transfers, I, equal A minus E.

4 Input transfers, J, equal B minus F.

5 Factor transfers, K, equal C minus G.

6 Net transfers, L, equal D minus H; L also equals I minus J minus K.

all four entries in the top row are measured in observed prices. This calculation begins with the construction of separate budgets

15

3

for farming, marketing, and processing (the principles and

practices of budget construction are detailed in Chapter

through 6). The components of these budgets are usually entered

in PAM as local currency per physical unit, although Lhe analysis

can also be clone using a foreign currency per unit.

The resulLs of private profitability calculations show the

ext. of actual competitiveness of the agricultural system,

given current technologies, output values, input costs, and

policy transfers. The normal cost of capital, defined as the

approximate minimum after-tax return that owners of capital

require to maintain the'.r investment in the system, is included

in domestic costs (C) ; hence, profits (D) are excess profits, or

,-bove-normal returns to operators of the activity. If private

pro'iLtability is negative (DO) are an indication of supernormal returns

and should lead to future increases of investment in the system,

if the farming area can be expanded and unless substitute crops

are more privately profitable.

Social Profitability

The second row of the accounting matrix contains social

prices. The term social refers to valuations that attempt to

measure comparative advantage or efficiency in the agricultural

production systems. In this context, efficient outcomes are

achieved when an economy's resources are used in activities that

create the highest levels of output and income. The PAM approach

measures the distoIrting effects of policies and market failures

that interfere with efficient outcomes. Social profitability,

defined as H, is an efficiency measure because outputs (E) and

inputs (F+G) are valued in prices that reflect scarcity values or

social opportunity costs. Social profit, like its private

analogue, is the difference between revenues and costs, all

measured in social prices (H=E-F-G).

16

The principles of social valuation and their empirical

application are explained in Chapter 6, and only the main

elements of the approach are summarized here. For outputs (E)

and inputs (F) that are traded internationally, the appropriate

social valuations are given by world prices---cif (costs,

insurance, freight) import prices for goods or services thiat are

imported or fob (free on board) export prices for exportables.

World prices represent the government's choice to import or

export or produce domestically; the social value of additional

domestic output is thus the foreign exchange saved by reducing

imports or earned by expanding exports, or for each unit of

production, the cif import or fob export price.

The services provided by the primary domestic factors of

production--labor, capital, and land--do not have world prices

because the markets for these services are domestic, not

international. The social valuation of each factor is found by

estimating the national income that is foregone because the

factor is not employed in its next best alternative use. For

example, if land is planted to wheat, that same land cannot grow

barley during the identical crop season; the social opportunity

cost of the land for the wheat system is thus the national income

lost because the land cannot produce barley in that crop year.

Similarly, the labor and capital used to produce wheat cannot

simultaneously provide services elsewhere in agriculture or in

other sectors of the economy. Social opportunity costs are

measured by the national income that is given up because

potential alternative activities are not pursued.

Social opportunity costs--or shadow prices--of primary

factors can only be approximated because of limitations imposed

by the availability of information. Specific techniques for

approaching this estimation are offered in Chapter 6. Because

domestic factors are used in all production pi'ocesses, the

difficult exercise of estimating shadow prices always arises.

Each PAM contains two cost columns, one for tradable inputs and

the other for domestic factors. Some domestic factors are used

17

directly in the production system; farmers, for example, employ their own and often hired labor, and their own equity as well as borrowed capital. The costs of these factors are entered in the

domestic factor column in PAM--private factor costs in element C and social factor costs in G.

Complications arise in PAM with handling intermediate inputs--materials, like fertilizer, pesticides, purchased seeds, compound feeds, electricity, transportation, and fuel. Many of these intermediates are traded internationally and thus have world prices to serve as social valuations. But some, such as electricity and transportation, are nontradable and hence do not have world prices. Even the tradable intermediates incur domestic marketing (handling and transportation) costs after importation or before exportation. The cif import or fob export prices are calculated at the port, whereas the relevant PAM prices need to apply to the location of production; hence, domestic marketing charges are added to cif import prices or subtracted from fob export prices to find social valuations

applicable to specific locations of agricultural systems.

Intermediate inputs are decomposed into the two cost categories of the accounting matrix, tradable inputs and domestic

factors. An example illustrates the process of decomposing intermediate goods or services. Fertilizer is usually a tradable intermediate input, if the country in question is a net importer of fertilizer, the social valuation of a specific kind of fertilizer for an agricultural system is given by the cif import price for that fertilizer plus the social costs of moving the input to the representative location of the system. Finding the import price is straightforward. The social valuation of the domestic marketing costs is different. It is necessary to study the transportation industry--road or rail--and disaggregate the costs into labor, capital, fuel, and so forth. The fuel costs, for example, then need to be further broken down through use of an appropriate world price and estimate of local transportation costs. (See Chapter 3 for an example of disaggregation.)

3.8

Effects of Divergences

The second identity, defintnq the accounting matrix, concerns the differences between private and social valuations of revenues, costs, and profits. Fcr each entry in the matrix--measured vertically, down the columns--any divergence

between the observed private (actual market) price and the estimated social (efficiency) price is explained 1y -he effects of policy or by the existence of market failures. This critical

relationship of policy analysis follows directly from the concept of social prices. To estimate social. prices, one corrects for the effects of distorting policies, those that lead to an inefficient use of resources and thus lower than potential levels .f output and income. Distorting polices are often introduced because decisionmakers are willing to accept some inefficiencies

(and consequent slower growth of income) to further nonefficiency objectives, such as redistribution of income or improved domestic food security. Assessing the tradeoffs between efficiency and

nonefficiency objectives is a central part of policy analysis.

Not all policies, however, distort the allocation of resources. Some policies are enacted expressly to improve i.2fficiency by correcting for the failures of product or factor

markets to operate properly. Market failures occur whenever

monopolies or monupsonies (seller or buyer control over market

prices), externalities (costs for which the imposer cannot be charged or benefits for which the provider cannot receive compensation), or factor market imperfections (inadequate

development of institutions to provide competitive services and full information) prevent a market from creating an efficient allocation of products or factors. The need arises to distinguish distorting policies, which cause losses of potential income, from efficient policies, which offset the effects of market failures and thus create greater income. Because

efficient policies correct, not cause, divergences, they reduce

the differences between private and social valuations.

19

Interpretation of the effects of divergences can be clarified by expanding the PAM to include six rows, as shown in

Table 2.2. In this expanded PA ,; each entry measuring the

effects of divergences (I, J, K, and L) is disaggregated into three categories--effects of market failures, effects of

distorting policies, and effects of efficient policies.

Additional efficient policies that offset market failures would

change the entries in the first and third rows. If market

failures are nonexistent in the product (output and tradable

input) markets, M' and N are zero and I and J are caused by

distorting policies. But if factor market imperfections exist

along with distorting factor policy, both 0 and S and possibly W

are positive components of K. The net transfer (L) is then made

up of the effects of distorting policy (I,.J, and the S part of

K) but also of the effects of factor market failures (the 0 part

of K) and of efficient policies offsetting them (the W part of

K). This situation is the one most commonly found in developing

countries.

In the absence of market failures aftecting the product

markets, all divergences between private and social prices of

tradable output and inputs are due to the effects of distorting

policy. Because the principles are identical for all tradable

products, the matrix entri,!s for tradable outputs (revenues) and

tradable inputs can be considered together. Output transfers

(I=A-E) and input transfers (J=B-F) arise from two kinds of

policies--commodity specific policies and exchange rate

policy--that cause divergences between observed and world product

prices.

20

Table 2.2 Expanded Policy Analysis Matrix

Revenues Costs Profits Tradable Domestic input factors

Private prices A B C D1 Social prices E F G H2 Effects of divergences 13 J4 K5 L6 and efficient policy Effects of market failures M N 0 P

Effects of distorting policy Q R S T Effects of efficient policy U V W X

1 Private profits, D, equals A minus B minus C

2 Social profits, H, equals E minus F minus G.

3 Output transfers, I, equal A minus E; I also equals M plus Q plus U.

4 Input transfers, J, equal B minus F; J also equals N plus R plus V.

5 Factor transfers, K, equal C minus G; K also equals 0

plus S plus W.

6 Net transfers, L, equal D minus H; L also equals I minus J minus K; L further equals P plus T plus X.

Policies that apply to specific commodities include taxes or

subsidies and trade restrictions. Producer revenues per unit can

be raised by producer subsidies (called deficiency payments in

agriculture), tariffs or import quotas on outputs (which raise

domestic prices), or domestic price supports enforced by

government stockpiling (which requires a complementary trade

restriction for tradable products). Commodity ;pecific policies

on inputs also affect private profitability. For example,

producer costs per unit can be lowered by direct input subsidies

or subsidies on imported inputs.

21

Exchange rate policies can also have an impact on product

prices. Typically, the PAM accounting is done in domestic currency, but world prices are quoted in foreign currency.

Hence, a foreign exchange rate is needed to convert world prices into domestic equivalents. (The analysis could instead be carried

out in foreign currency, but then an exchange rate would be

required to convert domestic factor costs from local currency

into that foreign currency.)

As explained in Chapter 6, the correct exchange rate to use

in converting world prices is one that reflects appropriate

macroeconomic policy. Overvaluation ot a foreign exchange rate

occurs when a government fails to adjust its exchange rate enough

to offset the effects of inflation cr of world price cnanges on international competitiveness. An overvalued exchange rate

creates an implicit tax on producers of tradable products because

too little domestic currency is earned by exports or paid for

imports. In the absence of commodity policy, the world price of

a tradable determines the domestic price for that good; when the

exchange rate is overvalued, the domestic price is lower than its

efficiancy level and domestic producers a-e effectively taxed by

this policy. Correction for this distortion in PAM is done by

converting world prices (E and F in the matrix) at the

appropriate exchange rate rather than at the official rate.

The social prices of domestic factors (G) are given by

determination of social opportunity costs, which reflect

underlying supply and demand conditions in domestic factor

markets. Factor prices are, therefore, inflienced by the

prevailing set of macroeconomic and commodity price policies.

The government can also enact tax or subsidy policies on one or

more of the factors (capital, labor, or land) which create a

divergence between private and social costs, resulting in a

subsidy to the system or a tax on the system. In addition,

market imperfections, arising from imperfect infcrmation or

underdeveloped institutional structures, are characteristic of

most developing countries and further influence factor prices.

22

The net transfer from policy and market failures (L) in is

found by summing the separate effects from the product and factor

markets (L=I-J--K). (Positive entries in the two cost categories,

J and K, represent negative transfers because private profits are

reduced by them, whereas negative entries in J and K represent positive transfers; hence, J and K are subtracted from I, a positive transfer, in calculating the net transfer, L.) The net

transfer from distorting policy is the sum of all. factor,

commodity: and exchange rate policies (apart from efficient

policies that offset market failures).

The net transfer is also found by comparing private and social profits. This transfer must by definition be identical in

the double entry accounting matrix (L=I-J-K) (D-H).

Disaggregation of the total net transfer shows whether each

distorting policy provides positive or negative transfers to the

system. The PAM approach thus permits comparison of the effects

of market failures and of distorting policies for the entire set

of commodity and macroeconomic (factor and exchange rate)

policies. This comparison can be made for the complete

agricultural system and separately for each of its outputs and

inputs. Two numerical examples of PAMs are presented in Box 2.1.

Comparisons Among Agricultural Systems

The entries in PAM allow comparisons among agricultural

systems that produce identical outputs, either within a single

country or across two or more countries. In the accounting

matrix, all measures are given as monetary units per physical

unit of some commodity. Comparisons can be drawn readily by

constructing PAM entries for two or more different systems that

produce the same quality of wheat (if necessary, premiums or

discounts can be used to correct for minor quality differences).

Further comparisons can be made between one country's wheat

systems and those in other wheat-producing countries; appropriate

exchange rates, incorporating corrections for differential

inflation not otherwise offset by exchange rate changes, are used

23

to convert the other countries' currencies into domestic

currency. If concern focuses solely on comparing one wheat system with another, the matrix entries provide all information

necessary for the analysis.

Comparisons between wheat and barley--or apples and

oranges--are another story. Appropriate ratios permit comparisons among systems producing different outputs. Both the numerator and the denominator of each ratio are entries in PAM defined in domestic currency units per physical unit of commodity. Therefore, the ratio itself is a pure number that is free of any commodity or monetary designation, because the currency per commodity unit labels cancel out. The task, then, is to define ratio indicators to substitute for the whole number

indicators in the PAM matrix. Illustrations of each of these indicators are presented in Chapter 7.

Private Profitability

For comparisons of systems producing identical outputs,

private profits (D=A-B-C) indicate competitiveness under existing policies. Construction of a ratio is required to permit

comparisons of the wheat systems with those producing other

commodities (or nonagricultural goods or services). Direct

inspection of the data for private profits is not sufficient. Profitability results are residuals and might have come from systems using very different levels of inputs to produce outputs

with widely varying prices. This difficulty might not be apparent in a wheat versus corn example, but it would arise in a

comparison of a wheat system with one producing a high value crop, such as strawberries, with a capital-intensive technology.

This ambiguity is inherent in comparing private profits of systems producing different commodities with differing capital

intensities.

The problem is circumvented by constructing a private cost ratio (PCR), defined as the ratio of domestic factor costs (C) to value added in private prices (A-B), that is, PCR = C/(A-B). Value added is the difference between the value of output and the

24

costs of tradable inputs; it shows how much the system can afford

to pay domestic factors (including a normal return to capital.)

and still remain competitive, that is, break even after earning

normal profits (where h-B-C=D=O). Clearly, the entrepreneurs in the system prefer to earn excess profits (D>O), and they can achieve this result if their private factor costs (C) are less

than their value added in privatr prices (A-B). They thus try to

minimize the private cost ratio by holding down factor and

tradable input costs to maximize excess profits.

Social Profitability

Social profits measure efficiency or comparative advantage.

To compare identical outputs, results can be taken directly from the second row of the PAM matrix--social profits equal social

revenues less social costs (H=E-F-G). When social profit is

negative, a system could not survive without the assistance of

policy. Such systems waste scarce resources by prcducing at social costs that exceed the costs of importing. The choice is clear for efficiency-minded economic planners: enact new

policies--or remove existing ones--to provide private incentives

for systems that generate social profits, subject to

nonefficiency objectives.

When systems producing different outputs are compared for

relative efficiency, a ratio is needed to compensate for the

problems of dissimilar commodities and technologies. The

domestic resource cost ratio (DRC), defined as G/(E-F), serves as

a proxy measure for social profits. No new information beyond

social revenues and costs is required to calculate a DRC. The

DRC plays the same substitute role for social profits as does the

PCR for private profits; in both instances, the ratio equals one

if its analogous profitability measure equals zero. Minimizing

the DRC is thus equivalent to maximizing social profits. In

cross-commodity comparisons, DRC ratios replace social profit

measures as indicators of relative degrees of efficiency.

25

Policy Transfers

Transfers are shown in the third row of the PAM. If market failures are unimportant, these transfers measure mainly the

effects of distorting policy. Efficient systems earn excess

profits without any help from the jovernment, and subsidizing

policy (L>O) substantially increases the final level o1 private

profits. Subsidizing policy is necessary to plLinit inefficient

systems to survive, but the consequent waste of resources needs

to be justified in terms of nonefficiency objectives.

Comparisons of the extent of policy transfers between two or more systems with different outputs also requires formation of

ratios (for reasons analogous to these offered above in The

discussions of private and social profits). The nominal

protection coefficient (NPC) is, a ratio that contrasts the

observed (private) commod jy price with a comparable world

(social) price, thereby indicating the impact of policy (and of

any market failures not corrected by efficient policy) causing a

divergence between the two kinds of prices. The NPC on tradable

outputs, defined as A/E, indicates the degree of output transfer;

an NPC of 1.10 shows that policies are driving up the actual

market price to a level 10 percent higher than the world price,

which would set the domestic price in the absence of policy.

Similarly, the NPC on tradable inputs, defined as B/F, shows the

degree of tradable input transfer. An NPC on inputs of 0.80

shows that policies are reducing input costs; the average market

prices for these inputs are only 80 percent of world prices,

their cost levels without policy.

The effective protection coefficient (EPC) is another

indicator of incentives. EPC is defined as the ratio of value

added in private prices (A-B) to value added in world prices

(E-F), or EPC = (A-B)/(E-F). This coefficient measures the degree of policy transfer from product market--output and

tradable input--policies, but it ignores the transfer effects of

factor market policies, Hence, EPC is not a complete indicator

of incentives.

26

An extension of the EPC to include factor transfers is the

profitability coefficient (PC), defined as PC=(A-B-C)/(E-F-G) or

D/H. The PC measures the incentive effects of all policies and

thus serves as a proxy for the net policy transfer (since L=D-H).

Its usefulness is restricted when either private or social

profits are negative because in these circumstances the signs of

both entries must be known to allow clear interpretation.

A final incentive indicator is the subsidy ratio to

producers (SRP), defined as SRP = L/E = (D-H)/E. The SRP shows the net policy transfer as a proportion of total social revenues,

or the proportion of revenues in world prices that would be required if a single subsidy or tax were to be substituted for

the entire set of actual commodity and macro policies. The SRP

permits comparisons of the extent to which policy subsidizes all

agricultural systems. The SRP measure can also be disaggregated

into component transfers to show separately the effects of

output, input, and factor policies.

Research Plan for Efficiency and Policy Analysis

There are usually six research inputs into the PAM

approach--private revenues (A), private costs of tradable inputs

(B), private costs of domestic factors (C), social revenues (E),

social costs of tradable inputs (F), and social costs of domestic

factors (G). Subtraction within the matrix then yields six

research results--private profits (D=A.-B-C), social profits

(H=E-F-G), output transfers (I=A-E), input transfers (J=B-F),

factor transfers (K=C-G), and net transfers (L=D-H=I-J-K). This

logic is correct. But the double-entry nature of the

matrix--which requires that the profits identity be satisfied for

calculations across the columns and that the effects of

divergences identity be met for calculations down the

rows--provides the analyst with some flexibility in research

strategy.

The goal is to find entries for all twelve elements in the

matrix. Gathering information on any six entries will permit

27

solution of the entire matrix (so long as no more than three entry inputs are in the same row and no more than two are in the

same column). In actual practice, filling in the twelve entries

in the accounting matrix is done pragmatically, using any information that is readily available. In almost every instance,

data on private revenues (A) and costs (B, C) are gathered first.

These data are usually observable, though rarely available in precisely the desired form (see Chapters 3 through 5). Once

private budgets arc complete, the calculation of private

profitability (D) is an easy first research output, completing

the entries in the first row of the matrix.

Completion of the second and third rows of each PAM is

usually pragmatic and rather ad hoc. The researcher takes

information from any available sources that are of reliably good

quality. Data are sought for three pairs of entries--social

revenues (E) and output transfers (I), social costs of tradable

inputs (F) and input transfers (J), and social costs of domestic

factors (G) and factor transfers (K).

Usually the world prices of outputs (E) are easier Io find

than are the transfer effects of output policy (I), especially

when these transfers are between consumers and agricultural

producers. But when the output transfers are in the form of

subsidies paid from the guvernment treasury, one occasionally has

better data on these subsidy payments than on the relevant world

prices. In this instance, it is preferable to enter the per unit

subsidy, the output transfer (I), as a research input and then to

calculate the implied world price as a residual (E=A-I).

Similarly, if better information exists on treasury costs of

fertilizer subsidies than on the appropriate world price for

fertilizer, the fertilizer part of J becomes data input anda

that of F a research output. Further, in the unlikely event that

an interest subsidy creates the entire divergence between private

and social costs of capital, data on the subsidy per unit could

be entered as the capital component of K and the shadow price of

capital needed for G could be found by subtraction (G=C-K), for

28

the capital component only. It is important to use the best quality information to fill in the matrix and not to worry about

whether components of E or I, F, or J, and G or K are research

inputs or results. In actual practice, therefore, social

valuations and measures of transfers are often arrived at in an

iterative fashion to incorporate all useful information.

Costs of Carrying Out the PAM Approach

The costs of applying the PAM method to a country's principal agricultural systems can vary widely. Primary

influences on research costs are: the prior availability of reliale micro data, especially at the farm level; the number and

complexity of representative agricultural systems selected for analysis; and the ease of verification of information and of gap-filling during field observations, in particular the time

required for administrative duties (such as making courtesy

calls, obtaining permissions, and clearing data) and logistics in

the field.

Application of the PAM approach is done most successfully if the research is led by an experienced, senior analyst who is able to devote full-time to the task. This senior analyst would

normally be aided by three or four younger assistants who have less experience and training and who, therefore, require direct

guidance to execute the field work properly. The economics of a

country's agricultural systems will be understood best if the

senior and junior personnel collaborate throughout all of the

steps of the PAM analysis, but especially during collection and

verification of data in the field.

Given a field circumstance of average availability of data, complexity of systems, and administrative hassle, a reasonable

estimate of the amount of time required of the senior researcher

for budget preparation is: selection of representative systems

(ranging between 6 and 20 systems), 1 man-month; verification of information in (20 or fewer) farm-level budgets, 3 man-months;

collection of survey information for (8 or fewer) post-farm

29

budgets, 1 man-month; and cleaning, checking and matching data

with the aid of a microcomputer, 1 man-month. These estimates

are based on the assumed availability of three or four full-time

junior assistants of average capability.

After budgets in private prices are completed, the costs of carrying out the social valuation depend on the prior

availability or ease of collection of comparable world prices,

adjusted to local wholesale markets, and on the prior

availability and ease of estimating appropriate shadow prices for

primary factors of production. In an average circumstance, two

man-months of experienced, senior time and six to eight of junior

time should be adequate for this task.

The interpretation and communication of results, for a

country of average complexity, should each require about two

senior man-months and six or eight junior ones. This estimate includes a full interpretation of results, including sensitivity

analyses with varying assumptions. Communication of the results

is assumed to comprise: writing a detailed technical paper on

issues, method, data, and results; constructing a careful policy

memo (in the manner suggested in Chapter 8); and preparing for

and delivering an oral presentation to policymakers and their

staffs.

The total senior research time that should be budgeted for carrying out a PAM analysis for a country of average difficulty

is thus about one man-year, including six man-months for

compiling budgets and six man-months for analyzing them, if the

effort also involves three or four man-years of research

assistance by junior personnel. This estimated requirement could

be reduced by no more than one-third for a country that has very

good data and working conditions, and the needed research time

might easily double in a country with modest data availability

and difficult circumstances for research.

30

Summary

The central purpose of PAM analysis is to measure the impact of government policy ov the private profitability of agricultural

systems and on the efficiency of resource use. Private

profitability and competitiveness are likely to be uppermost in

the minds of those concerned specifically with agricultural

incomes. Social profitability and efficiency is often emphasized by economic planners whose concern is with the allocation of

resources among sectors and the growth of aggregate income in the economy. Both sets of issues ultirmately focus on the incentive effects of policy--part of the difference between private and social profitability--and how such policy incenti-es might be

altered. Through evaluation of private and social revenues and costs, the PAM method is designed to illuminate these related

issues of agricultural policy analysis. The method is particularly well suited to empirical analysis of agricultural

price policy and farm incomes, public investment policy and

efficiency, and agricultural research policy and technological

change.

31

Box 2.1 Numerial Examples of Policy Analysis Matrices

Agricultural System A

Revenues Tradable

- Input Costs Domestic

- Factor Costs = Profits

Private prices 120 40 40 40

- Social prices 100 60 50 -10

= Effects of Policy 20 -20 -10 50

(20 percent (50 percent tariff on subsidy on output raises fertilizer cif price from reduces private 100 to cif costs from plus tariff 40 to 20; price of 120) other TICs=20)

(50 percent subsidy on credit reduces private costs from 20 to 10; other DFCs=30)

Therefore, policies create private profits (40) despite social profits (-10), because net policy transfer is 50.

Agricultural System B Tradable Domestic

Revenues - Input Costs - Factor Costs = Profits

Private prices 60 40 40 -20

- Social prices 120 60 50 10

= Effects of Policy -60 -20 -10 -30

(50 percent (50 percent (50 percent subidy to subsidy on subsidy on consumers of fertilizer credit reduces imports reduces private private costs reduces cif costs from from 20 to 10; price from 40 to 20; other DFCs=-30) 120 to dom- other TICs=20) estic price of 60)

Therefore, policies create negative private profits (and discourage production) (-20) despite positive social profits (10), because net policy transfer is -30.

32

Both examples show only the effects of distorting policies,

not of market failures. Fo: Agricultural System A, three

policies--a tariff on output, a subsidy on fertilizer, and a subsidy on credit--create positive transfers of 20, 20, and JO, respectively. The net -transfer, all from distorting policy in this example, is 50 (20 minus negative 20 minus negative 10). Distorting policy thus permits a socially unprofitable system (H

is negative 10) to operate with high private profits (D is 40) by

creating a positive policy transfer of 50 (D-H=L, or 40 minus

negative 10 equals 50). The second example, Agricultural System

B, shows the opposite result. Subsidies on tradable input- and

factors are the same as in the first example, but this system

receives an output price only half of the comparable world price.

A socially profitable system (H=10) is thus made privately

unprofitable (D=negative 20) because the net policy transfer is

negative (L=negative 60 minus negative 20 minus negative 10

equals negative 30; L also equals D minus H or negative 20 minus

10).

CHAPTER 3

CONSTRUCTING PAMs FOR COMMODITY SYSTEMS

This chapter presents a general framework for the

identification and development of representative agricultural

commodity systems. The first task for the analyst is to choose

systems that are closely related to the policy issues of

interest. In this identification process, decisions must be made

about all activities that comprise the commodity system. These

activities include farm production, movement of the commodity

from the farm to the processor, processing, and transport to a

wholesale market. By considering post-farm as well as the farm

activities, the analyst is able to assess all forces that

influence the efficiency and competitiveness of domestic

production.

_udgets of costs and returns provide the building blocks for

PAM, and the remainder of the chapter develops a format for the organization and presentation of budget data. The PAM uses both

private market and social prices for inputs and outputs. This

requirement forces the analyst to disaggregate cost and returns

information in two ways. First, quantity and unit price data are

usually necessary so that the analyst can apply social prices and

estimate social costs and returns. Second, costs are classified

into four categories: labor, capital, land, and tradable inputs.

Because the analyst needs to incorporate the impact of

divergences, private costs must be decomposed into the above

categories before social costs of production can be estimated.

Defining the Commodity System

Every economic activity is unique in some way. In the farm

sector, for example, commodity choices, land quality, and input

use patterns are almost never identical for any two farms.

Although it is possible to develop a different production model

for every farm in the agricultural sector, such exercises are

impractical because of limits on the resources available for

34

research. They are also useless for the policymaker, because the design of farm-specific policies is impossible. Instead, decisions affect broad categories of farmers, defined in terms of geographic location, commodity produced, and technologies. No two farmers are affected in exactly the same way by a particular policy action, and the policy-naker usually bases his decision on the "average" impact of the policy on some particular group of

farmers.

Developing a list of potential representative systems, and subsequent reduction of the list to a manageable number, thus

provide the initial task in the construction of PAM. This step is perhaps the most arbitrary, yet crucial, element of the PAM research. The analyst chooses characteristics that are similar across firms as the basis for the representative firm. Commodity produced, region of production, and production technology are perhaps the most common identification criteria. The choice of characteristic depends on the policy issue. If policy-makers are concerned about wheat price policy, the analyst will want to model a wheat commodity system. If the question is fertilizer pricing policy, the analyst will examine commodity systems that are prominent users of fertilizer. Because the change in policy could alter firm behavior, the analyst should also anticipate

commodity systems that potentially might become prominent, not

just those that are currently in operation.

While the analyst is searching for parameters that can aggregate individual farms into large "representative"

categories, concern must be given also to aggregate firms that are very different in other respects. The simultaneous concern for similarity and diversity means that more than one representative system may be necessary for PAM analysis. For example, if wheat farms have substantially different production technologies or soil qualities, a single commodity system is not representative of the wheat sector. One must distinguish representative wheat farms that are machinery-intensive, laboror animal-intensive, and utilize good or poor soils.

35

Recognizing diversity in representative systems is constrained by the resources and time available for :he research effort and by the differential impacts of policy on the systems. The preparation of budgets is a labor-intensive exercise, and projects rarely have the resources or time to study more than fifteen or twenty representative systems. At the same time, the effects of a particular policy (or set of policies) might not differ across systems in a major way. If changes in the price of wheat affect profitability similarly across different types of irrigation technologies, for example, little insight will be gained from explicit models of groundwater pumping and gravity-fed irrigation systems. More important distinctions might involve irrigation versus rainfed technology; the representative systems in this event would include groundwater pumping and nonirrigated wheat production techniques.

In most instances, the analyst will not know a priori the costs and returns of each potential system. Consequently, judgments about representativc systems will be arbitrary and reflect the ability of the analyst to anticipate the important and trivial differences in the results. As budgets are constructed and initial results evolve, diticrences between some representative system might prove to be small. If so, the representative system list can be shortened. Because the analyst does not want to leave out systems that are very different,

initial sets of representative systems should risk being too large rather than too small. Box 3.1 illustrates the system identification and selection process for wheat in Portugal.

The representative commodity system includes more than just a farm-level production activity. Consideration of only farm-level costs and returns would be sufficient to evaluate the efficiency and competitiveness of production for home consumption. But analysts usually will be interested in production for a domestic or foreign market that is geographically distinct from the farm. Selection aof

"representative" market destination means that post-farm costs

36

must be included in evaluation of the systems. For the purposes

of data collection and organization, the PAM framework defines a

commodity system to include four activities: farm production, delivery from farm to processor, processing, and delivery from processor to the wholesale market. Figure 3.1 illustrates the

structure of the commodity system model.

The critical consideration for activity selection is dictated by the requirements of the social evaluation exercise:

the domestically-produced product must be comparable to a commodity available on international markets. In some cases, the

activities of the representative system might not have cost or return entries. For example, both wheat flour and wheat grain

are available on world markets. As a result, analysts of

representative wheat systems may choose to ignore flour

processing altogether and emphasize system variations in wheat

production, transportation, and storage activities.

Alternatively, analytical interest might focus on wheat flour

processing. The farm production activity could be ignored

altogether, and analyses of flour production systems could

concentrate "cn variations in processing technologies, transportation and storage. In this example, wheat becomes a tradable input for the processing activity; its domestic market

price reflects the miller's costs, while the social

represented by the world market price plus the social

delivery to the mill.

value

costs

is

of

Classification of Inputs and Outputs

The budget of output revenues and input costs provides the

organizational framework for data collection. A budget is

constructed for each activity of the system. Data collection

begins with compilation of an inventory of inputs and outputs for

each activity. These items are categorized, quantified, and

priced in private and social terms. The costs and returns of

each activity are then summed to generate the total costs and

returns for the commodity system. (See Figure 3.2.)

37

FIGURE 3.1 The Structure of the Commodity System for PAM Analysis

FARM PRODUCTION Inputs and outputs for

production of raw materials. Evaluation stops at farmgate.

+

Move commodity from farm-gate FARM-TO-PROCESSOR to processing site. May include

storage and handling astransportation costs.

ACTIVITIES +

Processes commodity into PROCESSING consumer-acceptable form. May

involve physical transformation

or just packing, handling and quality control.

+

Move commodity from processing PROCESSING-TO- to market where domestic

WHOLESALE MARKET activity is comparable to tradable product. May include inputs and

outputs for farm-to-wholesale market if processing activity is irrelevant.

38

FIGURE 3.2 Input and output Categories for Activity Budgets

COSTS

I. Fixed Inputs

I1. Direct Labor

III. Intermediate Inputs

iv. Commodity-in-process

REVENUES

V. Outputs

39

Fixed Inputs

Budgets represent costs and returns on an annual (or single

crop) basis. However, fixed inputs have a useful life of many years, and only a portion of fixed input costs should be

attributed to a particular year's production. A simple approach

is to divide the cost of the caoital input by the useful life of

the input. But such calculations ignore the need for capital to

earn a rate of return on the investment. For example, if a wheat

farmer did not buy a tractor, the money could have been invested in some other activity on or off farm the farm. If this

potential investment could earn a positive rate of return, the

tractor investment must earn a return that is equal or higher.

The determination of an annual equivalent value for a fixed

input addresses the following question: "What is the annual

payment that will repay the cost of a fixed input over the useful

life of the input and in addition provide an economic rate of

return on the investment?" The mathematical formula for calculating the annual equivalent value is known as the capital

recovery factor. Lists of these factors for various interest

rates are found in most compounding and discounting tables.

The derivation of the formula for the capital recovery

factor can be illustrated in a few steps. A is defined as the

annual payment sufficient to repay the cost, Z, of the fixed

input, at the end of its useful life of n years. If one puts an

amount A into an investment earning a rate of return i, the total

value of the annual payments at the end of the fixed input's

useful life will be:

A[l + (14i) + (l+i) 2 +...+ (l+i) n -l] = Z (3.1)

where A(l+i)n- I represents the value of the initial deposit at -2the end of n years, the term A(l+i) n represents the value of

the second deposit at the end of n years, and so on until the

term A (1), which represents the value of the nth year payment.

The above formula calculates the amount necessary to repay

the cost of the fixed input. But if the fixed input is required

40

also to earn a positive rate of return, the value of the output produced by the fixed input is not just Z, but Z(l+i)n . Therefore, the annual cost equivalent calculation is expressed

as:

Afl(l+i)+(l+i) 2 +...+(l+i)n-l] = Z(l+i)n (3.2)

This expression can be altered by rearranging term